Journal of Substance Abuse Treatment xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Journal of Substance Abuse Treatment

Latent Class Analysis of Alcohol Treatment Utilization Patterns and 3-Year Alcohol Related Outcomes Orion Mowbray, Ph.D. a,⁎, Joseph E. Glass, Ph.D. b, Claudette L. Grinnell-Davis, Ph.D. c a b c

University of Georgia School of Social Work, 310 E. Campus Dr., Athens GA, 30602, USA University of Wisconsin-Madison, 1350 University Ave., Madison WI, 53706 USA University of Nebraska - Omaha, 6001 Dodge St. CPACS 206, Omaha, NE 68182, USA

a r t i c l e

i n f o

Article history: Received 4 June 2014 Received in revised form 28 January 2015 Accepted 29 January 2015 Available online xxxx Keywords: Treatment utilization Alcohol use disorders Latent class analysis

a b s t r a c t People who obtain treatment for alcohol use problems often utilize multiple sources of help. While prior studies have classified treatment use patterns for alcohol use, an empirical classification of these patterns is lacking. For the current study, we created an empirically derived classification of treatment use and described how these classifications were prospectively associated with alcohol-related outcomes. Our sample included 257 participants of the National Epidemiologic Survey on Alcohol and Related Conditions (NESARC) who first received alcohol treatment in the 3-year period prior to their baseline interview. We used latent class analysis to identify classes of treatment users based on their patterns of treatment use of 13 types of alcohol treatment. Regression models examined how classes of treatment use at baseline were associated with alcohol-related outcomes assessed at a 3year follow-up interview. Outcomes included a continuous measure of the quantity and frequency of alcohol use and DSM-IV alcohol use disorder status. Four classes of treatment users were identified: (1) multiservice users (8.7%), (2) private professional service users (32.8%), (3) alcoholics anonymous (AA) paired with specialty addiction service users (22.0%), and (4) users of AA alone (36.5%). Those who utilized AA paired with specialty addiction services had better outcomes compared to those who used AA alone. In addition to elucidating the most common treatment utilization patterns executed by people seeking help for their alcohol problems, the results from this study suggest that increased efforts may be needed to refer individuals across sectors of care to improve treatment outcomes. Published by Elsevier Inc.

1. Introduction Receiving treatment for alcohol-related problems is associated with reduced alcohol consumption (Dawson, Grant, Stinson, Chou, et al., 2006) and symptoms of alcohol use disorder (Dawson, Grant, Stinson, & Chou, 2006). While many individuals do not seek treatment for their alcohol-related problems (Cohen, Feinn, Arias, & Kranzler, 2007; Kessler et al., 1996), the receipt of any alcohol treatment is associated with higher odds of recovery (Weisner, Matzger, & Kaskutas, 2003). The array of available alcohol-related treatment services is vast. The treatment most frequently sought for alcohol-related problems is Alcoholics Anonymous (AA) (Cohen et al., 2007). Although the efficacy of AA represents a long-standing debate, current findings suggest that it is effective in reducing alcohol-related problems (Ferri, Amato, & Davoli, 2006; Kownacki & Shadish, 1999; Montgomery, Miller, & Scott Tonigan, 1995; Tonigan, Toscova, & Miller, 1996). The receipt of treatments delivered by persons or agencies specializing in addiction treatment is also common (Cohen et al., 2007). Generally, both inpatient (Keso & ⁎ Corresponding author at: University of Georgia School of Social Work, 310 East Campus Rd., Athens GA., 30602. Tel.: +1 706 542 5441. E-mail addresses: [email protected] (O. Mowbray), [email protected] (J.E. Glass), [email protected] (C.L. Grinnell-Davis).

Salaspuro, 1990; Waisberg, 1990) and outpatient (Bottlender & Soyka, 2005) approaches to treating alcohol use can be efficacious, with outpatient treatments being more cost effective (Weisner, Mertens, Parthasarathy, Moore, et al., 2000). Primary care providers, other professionals, and treatment agencies that do not specialize in the treatment of addiction are also within the broader sector of treatment for alcohol-related problems. For instance, alcohol screening and brief intervention has the potential to reduce unhealthy alcohol use in a variety of service settings (Amaro, Reed, Rowe, Picci, et al., 2010; Hermansson, Helander, Brandt, Huss, & Rönnberg, 2010; Schonfeld et al., 2010), especially among persons with less severe alcohol problems (Saitz, Horton, Sullivan, Moskowitz, & Samet, 2003). While it is common in alcohol research to operationalize treatment utilization as a binary event, such as the receipt of any alcohol treatment services versus none (Glass, Perron, Ilgen, Chermack, et al., 2010; Ilgen et al., 2011; Kessler et al., 1996; Mowbray, 2014), most individuals who seek help often report initiating multiple treatment episodes (SAMHSA, 2009) and seek treatment from a variety of service sectors (Cohen et al., 2007; Dawson, Goldstein, & Grant, 2012). To elucidate clinical and sociodemographic characteristics that predict the use of specific types of alcohol treatment, several prior studies have classified specific sources of treatment services into meaningful domains that pertain to the type of setting or format in which the treatment services were received

http://dx.doi.org/10.1016/j.jsat.2015.01.012 0740-5472/Published by Elsevier Inc.

Please cite this article as: Mowbray, O., et al., Latent Class Analysis of Alcohol Treatment Utilization Patterns and 3-Year Alcohol Related Outcomes, Journal of Substance Abuse Treatment (2015), http://dx.doi.org/10.1016/j.jsat.2015.01.012

2

O. Mowbray et al. / Journal of Substance Abuse Treatment xxx (2015) xxx–xxx

(Dawson et al., 2012; Hatzenbuehler, Keyes, Narrow, Grant, & Hasin, 2008; Kessler et al., 1996; Regier et al., 1993). For instance, Dawson et al. (2012) classified individuals' first utilization of alcohol treatment services hierarchically into mutually exclusive groups with respect to the setting of service delivery, including addiction specialty treatment settings, medical settings, non-medical settings, and no treatment, with specialty treatment taking precedence. Kessler et al. (1996) and Regier et al. (1993) used similar domains, but did not use mutually exclusive groupings. Another study classified treatments as either self-help, social services, alcohol or drug-specific services, and services received in emergency departments (Hatzenbuehler et al., 2008). These previous methods to classify alcohol treatment services into categories have several strengths. Researcher-specified domains of service use can be tailored to answer questions about specific types of treatment settings, and provide a more parsimonious analytic approach than examining each source of treatment separately. However, researcher-specified domains tend to have limited empirical justification and may not reflect how individuals generally use alcohol treatment services. For instance, while many individuals often attend AA and specialty addiction treatments concurrently (Gossop, Stewart, & Marsden, 2008; Timko, Moos, Finney, & Lesar, 2000) and cross-sector treatment referrals are common (Humphreys, 1997; Riordan & Walsh, 1994), prior studies have considered these treatments as existing in separate domains (e.g., medical versus non-medical) to explore predictors of their use separately (Dawson et al., 2012). Perhaps more importantly, a major limitation of the current literature on alcohol treatment utilization patterns is that research describing the outcomes associated with executed treatment utilization patterns is scant and limited to studies that have analyzed retrospective reports (Dawson, Grant, Stinson, & Chou, 2006; Dawson, Grant, Stinson, Chou, et al., 2006). To our knowledge, no studies have examined outcomes associated with alcohol treatment utilization patterns with longitudinal data from the general population. In the statistical and substantive literature, person-centered methods have become increasingly popular (Collins & Lanza, 2013), which have the ability to generate respondent profiles based on their prior utilization of specific sources of alcohol treatment. Through the use of person-centered methods, distinct patterns of treatment utilization can be established that are driven by an empirical classification of participant data and may more accurately reflect the actual histories of those who have been treated for alcohol-related problems. The current study had two aims. First, we established empirically derived treatment utilization profiles of individuals with alcohol use problems to provide a better understanding of alcohol treatment utilization patterns in the general population. Second, we examined whether these utilization profiles were associated with differences in alcohol-related outcomes 3 years later. To accomplish these aims, we applied latent class analysis (LCA) in a longitudinal nationally representative sample of adults who sought treatment for alcohol use over a 3-year period to construct profiles of participants' utilization of thirteen different sources of alcohol treatment. Next, we examined whether these profiles of treatment utilization predicted levels of alcohol consumption, abstinence from alcohol, and the presence of alcohol use disorder 3 years later. 2. Materials and methods 2.1. Sample Data are from waves 1 (2001–2002) and 2 (2004–2005) of the National Epidemiologic Survey on Alcohol Related Conditions (NESARC). NESARC is a population-representative survey of United States adults aged 18 or older living in noninstitutionalized settings (Grant, Kaplan, & Stinson, 2007; Grant, Moore, Shepard, & Kaplan, 2003; Hasin, Stinson, Ogburn, & Grant, 2007). The NESARC data are weighted to represent the U.S. general population based on the 2000 decennial census and to reflect survey design characteristics including oversampling of women,

Black and Hispanic individuals, and persons of younger age (Grant et al., 2003). Our analytic sample included NESARC participants who reported their first episode of treatment for alcohol related problems no more than 3 years prior to wave 1 of the NESARC (n = 257). A 3-year time window was chosen to create roughly equivalent measurement periods for the baseline and follow-up measures, and also to reduce the amount of time between the exposure (treatment utilization) and outcome variables. The presence of prior treatment utilization was established at wave 1 through the question “Have you ever gone anywhere or seen anyone for a reason that was related in any way to your drinking….” For participants that affirmed, they were asked “How old were you the first time you went anywhere or saw anyone for help with your drinking?” We calculated the number of years since the participant's first treatment episode to establish parameters of the current sample (i.e., first utilized treatment no more than 3 years before the baseline interview). The University of Georgia Institutional Review Board approved this research. 2.2. Measures 2.2.1. Use of specific sources of alcohol treatment Participants were queried about their use of 13 different sources of alcohol treatment services. The 13 sources of treatment services assessed included AA, family services or other social service agencies, alcohol detoxification services, inpatient programs, outpatient programs, alcohol rehab programs, emergency rooms, halfway houses, crisis centers, employee assistance programs, clergy/priest/rabbi, private professionals (private physician, psychiatrist, psychologist, social worker, or any other professional), and “other” services. Participant responses were dichotomous (i.e., used the service versus not) for each source of treatment. Although the NESARC treatment utilization instrument has not been validated against “gold standards” of treatment utilization such as medical records (Glass & Bucholz, 2011; Killeen, Brady, Gold, Tyson, & Simpson, 2004), it has been used in a number of landmark studies on alcohol treatment utilization (Cohen et al., 2007; Dawson, Goldstein, & Grant, 2007). 2.2.2. Sociodemographic characteristics Sociodemographic variables included gender, race/ethnicity (categorized as White, not Hispanic; Black, not Hispanic, and other), age (categorized as 18–34, 35–55 and 55 or over), and annual household income ($0–19,999, $20,000–34,999, $35,000–69,999 and $70,000 or more). 2.2.3. Clinical variables At wave 1, participants were classified as having a lifetime DSM-IV drug use disorder, including abuse or dependence of cannabis, cocaine or crack, tranquilizers, inhalants, stimulants, opioid painkillers, heroin, other prescription drugs, hallucinogens, and sedatives. Participants were also assessed for the presence of a lifetime DSM-IV diagnosis of substance-unrelated psychiatric disorder (i.e., “mental health” disorder), including major depressive disorder, mania, dysthymia, hypomania, panic disorder, any anxiety disorder, posttraumatic stress disorder, or any personality disorder. Both the drug use disorder and mental health disorder measures were dichotomized into any disorder versus none. Additionally, a count of lifetime symptoms of DSM-IV alcohol dependence and alcohol abuse (range 0–11) was constructed for each participant using responses to the symptom measures of the Alcohol Use Disorder and Associated Disabilities Interview Schedule (AUDADIS-IV) (Grant et al., 2007; Ruan, Goldstein, Chou, Smith, et al., 2008). Age of alcohol use disorder onset was assessed with the question, “about how old were you the first time some of these experiences began to happen around the same time,” where “these experiences” referred to alcohol use disorder diagnostic criteria assessed by the AUDADIS-IV. For a measure of alcohol consumption, we calculated Alcohol Use Disorders Identification Test-Consumption (AUDIT-C; range 0–12) scores from the W1 data, which is a continuous measure of the quantity

Please cite this article as: Mowbray, O., et al., Latent Class Analysis of Alcohol Treatment Utilization Patterns and 3-Year Alcohol Related Outcomes, Journal of Substance Abuse Treatment (2015), http://dx.doi.org/10.1016/j.jsat.2015.01.012

O. Mowbray et al. / Journal of Substance Abuse Treatment xxx (2015) xxx–xxx

and frequency of alcohol use used to identify potential hazardous drinking in the past-year (Bradley, DeBenedetti, Volk, Williams, et al., 2007; Bush, Kivlahan, McDonell, Fihn, & Bradley, 1998; Dawson, Grant, Stinson, & Zhou, 2005). 2.2.4. Alcohol-related outcome variables For a continuous measure of alcohol consumption at follow-up, AUDIT-C scores were calculated from the wave 2 data. Also, participants self-identified as abstinent from alcohol for the duration of the previous year, or not, at wave 2. Finally, participants were classified as having any past-year alcohol use disorder or no past-year alcohol use disorder at wave 2 using the AUDADIS-IV diagnostic measures. 2.3. Latent class analysis LCA was used to assign participants into groups based upon their reported patterns of treatment utilization. As opposed to exploratory and confirmatory factor analysis, which creates a taxonomy of related variables, LCA creates a taxonomy of people with similar response sets that can be used to describe related patterns of events (Roesch, Villodas, & Villodas, 2010). A detailed overview of LCA methodology is beyond the scope of this study and is available elsewhere. To identify a latent class model that best represented the number and nature of treatment utilization profiles, we used methods widely recommended in the statistical literature (Nylund, Asparouhov, & Muthén, 2007). There is no current consensus regarding a single statistical measure that identifies the best number of classes in any given population (Nylund et al., 2007). Rather, it is preferred to examine class solutions across a number of measures of statistical fit, considering changes in fit as the number of classes included in the solution increases. With this in mind, a series of models was examined, each model with an increasingly higher number of classes specified. The final model selection used to establish the appropriate number of classes was based on both conceptual considerations and several statistical fit indices, including the Akaike Information Criterion (AIC) (Akaike, 1974), Bayesian Information criterion (BIC) (Schwarz, 1978), sample size adjusted BIC (SSABIC) (Sclove, 1987), Lo–Mendell–Rubin likelihood ratio test (LMR-LRT) (Lo, Mendell, & Rubin, 2001), and an entropy measure (Ramaswamy, Desarbo, Reibstein, & Robinson, 1993). Latent class analyses were completed using Mplus version 7.0, which can accommodate the complex survey design methodology of NESARC through the use of strata, cluster, and weight variables provided in the data (Muthén & Muthén, 2012). Additionally, methods to generate accurate subpopulation estimates were employed in the LCA (Asparouhov, 2005; Kaplan & Ferguson, 1999). The LCA employed maximum-likelihood estimation with robust standard errors (MLR). Models were run with 200 random sets of starting values and 20 final stage optimizations, in order to avoid the issue of local maxima and to ensure that all values converged on identical solutions. 2.4. Three-year analysis of alcohol related outcomes After a best fitting latent class solution was identified, assigned class memberships for each participant were exported as a categorical variable to be used in regression analyses in Stata version 13 (StataCorp, 2013) to predict 3-year alcohol-related outcomes. Linear regression was used to model AUDIT-C scores, and logistic regression was used to model alcohol abstinence and alcohol use disorder status as a function of latent class membership, controlling for sociodemographic characteristics, lifetime drug use disorder, lifetime mental health disorder, and age of alcohol use disorder onset. In order to control for unmeasured stable between-person differences, we controlled for AUDIT-C scores and AUD symptom counts at the baseline interview. All regression analyses adjusted for the complex survey design of NESARC and used methods to generate accurate subpopulation estimates.

3

3. Results 3.1. Sample characteristics Sample descriptive characteristics of NESARC respondents who reported a first-time treatment episode no more than 3 years prior to wave 1 of the NESARC are displayed in Table 1. The sample was predominantly male (73.0%), aged 18 to 34 (57.4%), and was comprised mainly of persons who were White, not Hispanic (72.9%). The most common annual household income category was $35,000–$69,999 (32.3%). Approximately 23% met criteria for a lifetime drug use disorder, and 56% reported any lifetime substance-unrelated psychiatric disorder. Participants averaged 7.0 alcohol use disorder symptoms at wave 1, reported an average age of alcohol use disorder onset at 29.3 years, and an average AUDIT-C score of 5.7. At the follow-up interview, about 24% of participants were abstinent from alcohol use. The average AUDIT-C score at follow-up was 4.5, and 44.8% reported a past year alcohol use disorder. In terms of treatment utilization at baseline, AA (61.3%), alcohol or drug rehabilitation programs (36.3%), and treatment by private physician, psychiatrist, psychologist, social worker, or any other professionals (37.2%) were the most common forms of treatment (see Table 2). 3.2. Estimation of latent classes Five latent class models were fitted to the data, beginning with a one-class solution. The number of random sets of starting values and final stage optimizations were increased to 600 and 30 for the fiveclass model to achieve convergence. Based on a comparison of model fit indices across the LCA models (see Table 3) and our substantive interpretation, we deemed that the data were best represented by a fourclass solution. The AIC and SSABIC were lowest in the 3-class model, but changes in fit were minor when comparing the four-class and five-class models. The Lo–Mendell–Rubin Likelihood Ratio identified the four-class model as a more parsimonious representation to the data than a five-class model. The two-class model had the best entropy value, but overall entropy statistic values were similar (range Table 1 Sample characteristics of NESARC respondents who first sought help for drinking up to 3 years prior to the wave 1 interview. N = 257 Age (in years) 18–34 34–55 55 and over Race/Ethnicity White, Not Hispanic Black, Not Hispanic Other Gender (male) Annual household income $0 to $19,999 $20,000–$34,999 $35,000–$69,999 $70,000 or more Lifetime substance use diagnosis⁎ Lifetime MH diagnosis† Wave 1 lifetime AUD symptom count Age of AUD onset Wave 1 AUDIT-C score Wave 2 AUDIT-C score Wave 2 alcohol abstinence Wave 2 past year AUD status

Weighted % or (M)

SE

57.4 36.2 6.4

0.035 0.034 0.017

72.9 8.3 18.8 73.0

0.036 0.017 0.017 0.032

21.7 25.6 32.3 20.4 23.4 56.2 (6.99) 29.3 (5.73) (4.50) 24.4 44.8

0.029 0.035 0.037 0.032 0.030 0.039 0.204 0.994 0.267 0.282 0.029 0.038

MH = mental health, AUD = alcohol use disorder, AUDIT-C = Alcohol Use Disorders Identification Test-Consumption. ‡ Lifetime DSM-IV diagnosis of any mental illness, including major depressive disorder, mania, dysthymia, hypomania, panic disorder, any. ⁎ Lifetime DSM-IV criteria for abuse or dependence for marijuana, cocaine or crack, tranquilizers, stimulants, painkillers, other prescription drugs, hallucinogens, and sedatives. † Mood, anxiety, or personality disorder.

Please cite this article as: Mowbray, O., et al., Latent Class Analysis of Alcohol Treatment Utilization Patterns and 3-Year Alcohol Related Outcomes, Journal of Substance Abuse Treatment (2015), http://dx.doi.org/10.1016/j.jsat.2015.01.012

4

O. Mowbray et al. / Journal of Substance Abuse Treatment xxx (2015) xxx–xxx

Table 2 Thirteen types of alcohol treatments among NESARC respondents who sought help for drinking up to 3 years prior to the wave 1 interview. N = 257

Weighted SE %

Alcoholics Anonymous, Narcotics or Cocaine Anonymous Meeting, or any 12-step meeting? Family services or another social service agency? Alcohol or drug detoxification ward or clinic? Inpatient ward of psychiatric or general hospital or community mental health program? Outpatient clinic, including outreach programs and day or partial patient programs? Alcohol or drug rehabilitation program? Emergency room for any reason related to your drinking? Halfway house or therapeutic community? Crisis center for any reason related to your drinking? Employee assistance program (EAP)? Clergyman, priest, rabbi, or any type of religious counselor for any reason related to your drinking? Private physician, psychiatrist, psychologist, social worker, or any other professional? Other agency or professional?

61.3

0.039

17.7 23.3 11.9

0.031 0.029 0.022

26.3

0.033

36.6 15.8 2.9 1.2 6.5 9.2

0.032 0.026 0.001 0.005 0.022 0.021

37.2

0.039

13.1

0.024

0.79–0.83) and offered little distinction among the classes. Compared to the four-class model, which we describe in the next section, the fiveclass model (not shown) added a relatively small class (5% of the sample) that had a high probability (0.66) of “other treatment” (which was not substantively informative), and none of the remaining treatment variables made this class distinct from models with fewer classes.

Table 4 Estimated posterior probabilities of obtaining the 13 types of treatment for alcohol use. N = 257

Multiservice use (8.7%)

AA alone (36.5%)

AA combined with specialty (22.0%)

Private professionals (32.8%)

AA FS Detox Inpatient Outpatient Rehab ER Halfway Crisis EAP Clergy Private Other

0.83 0.42 0.84 1.00 0.72 0.76 0.56 0.09 0.11 0.00 0.37 0.95 0.17

0.86 0.05 0.06 0.02 0.15 0.24 0.11 0.00 0.00 0.03 0.02 0.14 0.09

0.88 0.18 0.58 0.08 0.53 0.91 0.18 0.09 0.01 0.09 0.05 0.35 0.13

0.0 0.28 0.02 0.03 0.07 0.00 0.09 0.00 0.00 0.11 0.14 0.53 0.18

Note: AA = Alcoholics Anonymous. FS = family or social service settings. DEtox = alcohol or drug detoxification ward or clinic. Inpatient = inpatient ward of psychiatric or general hospital or community mental health program. Outpatient = outpatient clinic, including outreach programs and day or partial patient programs. Rehab = alcohol or drug rehabilitation program ER = emergency room. Halfway = halfway house or therapeutic community. Crisis = crisis center. EAP = employee assistance program. Clergy = clergyman, priest, rabbi, or any type of religious counselor. Private = private physician, psychiatrist, psychologist, social worker, or any other professional. Other = other agency or professional. Probabilities greater than 0.4 are displayed in bold to assist with interpretation.

“private professional service users” class. In all classes, probabilities of attending halfway houses, crisis centers, employee assistance programs, clergy/priest/rabbi, and “other treatments” for drinking were relatively low (range 0.00–0.24).

3.3. Treatment utilization profiles

3.4. Class membership as a predictor of 3-year alcohol related outcomes

Table 4 contains the estimated posterior probabilities of endorsing the 13 alcohol treatment sources conditional upon latent class membership for the four-class model. Fig. 1 contains a profile plot to assist with interpretation. Latent class 1 comprised 8.7% of the sample, which had high probabilities for endorsing a multitude of treatment sources, including AA (0.83), family services or other social service agency (0.42), alcohol detox services (0.84), inpatient programs (1.0), outpatient programs (0.72), alcohol rehab programs (0.76), emergency rooms (0.56) and private professionals (0.95). This class is best characterized as the multiservice users class. Latent class 2 comprised 36.5% of the sample, which had the highest probability of attending AA (0.86) and very low probabilities for all other services (0.00–0.24). We deemed this class to be the users of AA alone class. Latent class 3 comprised 22.0% of the sample, which had a higher probability of endorsing AA (0.88), alcohol detox services (0.58), outpatient programs (0.53) and alcohol rehab programs (0.91). This class is best characterized as an AA paired with specialty addiction service users class. Latent class 4 comprised 32.8% of the sample, which had a high probability of attending treatment with private professionals (0.53), and relatively low probabilities of all other treatments. As such, this class is best characterized as a

A linear regression model examining the association between class membership and AUDIT-C scores at follow up, controlling for sociodemographic and clinical characteristics is presented in Table 5. With the users of AA alone class as the reference group (the class with the highest prevalence), individuals in the AA paired with specialty addiction service users class had significantly lower AUDIT-C scores at follow up (B = −2.13, 95% CI = −3.95 to −0.59). Higher wave 1 AUDIT-C scores, age (55 and over), gender, and age of onset were also significantly associated with follow up AUDIT-C scores. Furthermore, when using AA paired with specialty addiction service users class as the reference group (not shown in table), individuals in the private professional service users class were more likely to have higher AUDIT-C scores at follow up (B = 2.40, 95% CI = 0.85–3.94). The logistic regression model examining alcohol abstinence at follow up showed that individuals in the multiservice users class had higher odds of abstinence compared to individuals in the users of AA alone class (OR = 4.79, 95% CI = 1.03–22.32). Also, individuals in the AA paired with specialty addiction service users class had higher odds of abstinence than individuals in the users of AA alone class (OR = 14.26, 95% CI = 3.98–51.08). The presence of a lifetime substance use diagnosis and lower AUDIT-C score were both associated with lower odds of alcohol abstinence at follow up. When using the multiservice users class as the reference group (not shown in table), individuals in the AA paired with specialty addiction service users class had higher odds of abstinence (OR = 3.44, 95% CI = 1.03–11.54), whereas individuals from the private professional service users class showed lower odds of abstinence at follow up (OR = 0.11, 95% CI = 0.02–0.57). The logistic regression model examining alcohol use disorder status at follow up showed that individuals in the AA paired with specialty addiction service users class had significantly lower odds of any alcohol use disorder compared to individuals in the users of AA alone class (OR = 0.26, 95% CI = 0.07–0.96). Wave 1 AUDIT-C scores, age, gender and age of onset were also significantly related to follow-up alcohol use disorder status. When using the AA paired with specialty addiction service users class as reference (not shown in table), individuals in the

Table 3 Fit indices for one-class to five-class latent class analysis models of alcohol treatment utilization. Model

1 class 2 class 3 class 4 class 5 class

Fit statistics AIC

BIC

SSABIC

Entropy

LRT-LMR

2,857.66 2,671.40 2,619.33 2,610.96 2,611.13

2,903.80 2,767.22 2,764.84 2,806.16 2,856.01

2,903.80 2,681.63 2,634.86 2,631.79 2,637.26

– 0.79 0.83 0.80 0.82

– p b .01 p b .01 p b .01 ns

Note. AIC, Akaike Information Criteria; BIC, Bayesian Information Criteria; SSABIC, SampleSize Adjusted Bayesian Information Criteria; LRT-LMR, p-value for the Lo–Mendell–Rubin Likelihood Ratio Test. Selected model is in bold type.

Please cite this article as: Mowbray, O., et al., Latent Class Analysis of Alcohol Treatment Utilization Patterns and 3-Year Alcohol Related Outcomes, Journal of Substance Abuse Treatment (2015), http://dx.doi.org/10.1016/j.jsat.2015.01.012

O. Mowbray et al. / Journal of Substance Abuse Treatment xxx (2015) xxx–xxx

5

Latent class profile plot of treatment types 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

Multiservice users (8.7%)

Private professional service users (32.8%)

AA paired with specialty addiction service users (22.0%)

Users of AA alone (36.5%)

Fig. 1. Latent class profile plot of treatment types.

private professional service users class showed higher odds of any alcohol use disorder (OR = 4.86, 95% CI = 1.35–17.59). Finally, no significant differences emerged when examining class membership and wave 1 AUDITC scores, controlling for sociodemographic and clinical characteristics. 3.5. Sensitivity analyses Our LCA (see Section 3.3) only included participants who first received treatment in the 3 years prior to the baseline interview to reduce

the amount of time between the exposure (treatment utilization) and outcome variables. Thus, to evaluate the robustness of the latent class profiles, we replicated the LCA among those who had ever sought treatment in their lifetime (n = 1,247) based on reports at the baseline and follow-up interviews. The substantive interpretation of the latent class profiles remained unchanged (see Appendix A for estimated posterior probabilities). We note that in the sensitivity analysis, the latent class prevalences for the AA alone class (24.4%) and private professional class (28.6%) were slightly smaller, whereas the multiservice use class

Table 5 Multivariate regression models predicting 3 year alcohol-related outcomes. N = 250

Class membership AA Alone Multiservice use AA with specialty services Private professionals Age (in years) 18–34 34–55 55 and over Race/Ethnicity White Black Other Gender (male) Annual household income $0 to $19,999 $20,000–$34,999 $35,000–$69,999 $70,000 or more Lifetime substance use diagnosis2 Lifetime MH diagnosis3 Age of AUD onset Wave 1 AUDIT-C score Wave 1 AUD symptom count

AUDIT-C

Alcohol abstinence

AUD status

B

95% CI

OR

95% CI

OR

95% CI

– −1.34 −2.13⁎⁎ 0.48

– −3.28 to 0.48 −3.95 to −0.59 −1.17 to 1.43

4.79⁎ 14.26⁎⁎ 0.52

1.03 to 22.32 3.98 to 51.08 0.13 to 2.02

– 0.86 0.26⁎ 1.28

– 0.17 to 4.21 0.07 to 0.96 0.39 to 4.16

– −0.16 −3.42⁎⁎

– −1.72 to 1.31 −6.35 to −1.19

– 0.72 2.92

– 0.21 to 2.44 0.29 to 29.43

0.17⁎⁎ 0.01⁎

0.03 to 0.86 0.01 to 0.01

– −1.01 1.08 2.28⁎⁎

– −2.67 to 0.51 −0.29 to 2.41 0.93 to 3.58

2.67 0.67 0.49

0.72 to 9.79 0.19 to 2.36 0.18 to 1.38

– 0.45 1.75 3.34

– 0.11 to 1.97 0.57 to 5.27 1.33 to 8.43

– 0.19 −1.29 −0.51 1.29 0.57 0.07⁎ 0.31⁎⁎ 0.05

– −1/26 to 1.63 −2.73 to 0.15 −2.23 to 1.19 −0.22 to 2.52 −0.57 to 1.97 0.02 to 0.15 0.13 to 0.45 −0.14 to 0.23

– 0.86 2.85 1.50 0.26⁎ 0.63 0.98 0.80⁎⁎ 0.92

– 0.22 to 3.45 0.89 to 9.03 0.43 to 5.21 0.08 to 0.81 0.21 to 1.86 0.93 to 1.02 0.70 to 0.92 0.75 to 1.12

– 0.43 0.43 1.25 1.23 1.15 1.21⁎⁎ 1.23⁎⁎ 0.88

– 0.09 to 1.87 0.11 to 1.74 0.29 to 5.22 0.38 to 3.92 0.39 to 3.36 1.11 to 1.31 1.08 to 1.41 0.39 to 4.17

1 Classsfied as having any insurance at wave 1 of the NESARC. Abstinence: AA alone class n = 203, multiservice use class n = 94, AA with specialty services class n = 214, private professionals class n = 54. ⁎ p b .05. ⁎⁎ p b .01. 2 Lifetime DSM-IV criteria for abuse or dependence for marijuana, cocaine or crack, tranquilizers, stimulants, painkillers, other prescription drugs, hallucinogens, and sedatives. 3 Lifetime DSM-IV diagnosis of any substance-unrelated psychiatric disorder, including major depressive disorder, mania, dysthymia, hypomania, panic disorder, any anxiety disorder, posttraumatic stress disorder, or any personality disorder.

Please cite this article as: Mowbray, O., et al., Latent Class Analysis of Alcohol Treatment Utilization Patterns and 3-Year Alcohol Related Outcomes, Journal of Substance Abuse Treatment (2015), http://dx.doi.org/10.1016/j.jsat.2015.01.012

6

O. Mowbray et al. / Journal of Substance Abuse Treatment xxx (2015) xxx–xxx

(12.1%) and AA combined with specialty services class (34.9%) were slightly larger. These patterns may reflect that participants had more time to use multiple services, and thus slightly more people would fall into classes that were comprised of multiple treatment types. In our regression analyses (see Section 3.4), controlling for baseline alcohol consumption allowed participants to serve as their own controls, but changes over time in participant characteristics could have biased the results. Hence, we specified an additional model that adjusted for values of income, drug use disorder, and mental health disorder at the follow-up interview. The directions of the relationships and patterns of statistical significance remained the same (results are available from the first author upon request). 4. Discussion The first aim of this study was to establish empirically derived treatment utilization profiles of individuals with alcohol use problems in order to provide a better understanding of alcohol treatment utilization patterns. Based on an LCA of NESARC respondents who sought treatment for the first time in the three 3 years prior to their baseline assessment, individuals who used treatment for alcohol use disorders were best classified into four profiles based on their service use patterns: a multiservice user class, a private professional service user class, an alcoholics anonymous (AA) paired with specialty addiction service user class, and a users of AA alone class. These profiles provide a novel person-centered classification of alcohol treatment utilization. As opposed to previous researcher-driven classification systems, these results more accurately reflect the lived treatment experiences of those who have been treated for alcohol-related problems. Specifically, this work departs from previous attempts to categorize alcohol treatment utilization in both its methodological approach and substantive findings. Previous work has suggested that AA should be grouped into a “nonmedical service” category to include 12-step treatment for alcohol problems within the same domain as treatment from human services sectors (e.g. social service agencies, crisis centers) (Dawson et al., 2012). In contrast, the profiles of service use found in current study suggested that AA may be better characterized in a class on its own or considered along with specialty addiction treatment services. We also did not find support for a “medical service” profile of treatment users (i.e., users of private physicians, others professionals, and emergency departments) (Dawson et al., 2012). Rather, persons who used private professionals appeared to most commonly use those services on their own, whereas users of emergency departments for alcohol-related services fell into a multiservice use class. This may suggest that emergency department treatment of alcohol problems is most common among those with a higher problem severity or a propensity to seek treatment from multiple sources, rather than a preference to seek treatment from medical providers. While these findings do not perfectly align with previous efforts to classify treatment use among individuals with alcohol use problems, there were some similarities. For instance, most prior studies have emphasized theoretical distinctions between specialty and non-specialty care (Dawson et al., 2012; Kessler et al., 1996; Regier et al., 1993), which was supported in part in the current analysis such that a profile emerged that consisted of service users who attended alcohol-specific services, such as outpatient programs, rehabilitation programs, and AA. Perhaps, a “non-specialty” or “informal” class of treatment users did not emerge on its own in our analyses because these treatments are less frequently sought in the general population. The second aim of our study also yielded important findings. The early career treatment utilization profiles identified in this study were associated with differences in alcohol consumption and the presence of alcohol use disorder 3 years later. When examining AUDIT-C scores, alcohol abstinence, and alcohol use disorder status at the 3-year follow-up interview, individuals from the AA paired with specialty addiction service users class showed lower AUDIT-C scores, higher odds of abstinence, and a lower odds of any alcohol use disorder compared

to individuals from the class who used AA alone. This extends findings from prior clinical and cross-sectional general population studies, which have demonstrated support for the use of both AA and formal treatment over using formal treatment alone (Dawson, Grant, Stinson, & Chou, 2006; Dawson, Grant, Stinson, Chou, et al., 2006; Gossop, Harris, Best, Man, et al., 2003). Together, our findings and findings from prior studies suggest that specific treatment utilization patterns may produce differential alcohol-related outcomes, especially among individuals with no prior treatment episodes. While referrals between AA and specialty treatment services are common (Fiorentine, 1999; Kelly, Stout, Zywiak, & Schneider, 2006), formal efforts may be needed to encourage AA programs and specialty treatment providers to increasingly conduct cross-sector treatment referrals. Moreover, given that the use of AA alone was the most common pattern of treatment seeking, efforts directed towards encouraging AA members (e.g., sponsors) to refer new members to specialty forms of addiction treatment may be indicated.

4.1. Limitations While this study offers several strengths in its presentation, including the use of a large, longitudinal, and nationally representative dataset of individuals with alcohol problems in the United States, there are several limitations. While the benefit of limiting the analytic sample to those who sought treatment for the first time in the 3 years prior to baseline created a more homogenous sample and reduced the amount of time between the initial exposure to alcohol treatment and the assessment of subsequent alcohol-related outcomes, this approach also reduced the sample size. Hence, we were not able to conduct comparisons of subgroups within this sample, such as stratifying the LCA by DSM-IV alcohol abuse versus dependence status. Additionally, while wave 1 alcohol use consumption and alcohol use disorder symptoms were controlled for in all regression models, we did not control for factors such as treatment utilization between waves 1 and 2 of the NESARC that may have influenced outcomes 3 years later (Mowbray, 2014; Tonigan et al., 1996). However, the strength of our approach was it offered a clean way to estimate how initial treatment utilization patterns predict subsequent alcohol outcomes. Future studies may wish to conduct within-person analyses to estimate how changes in treatment utilization are associated with changes in treatment outcomes. Moreover, the treatment utilization instruments assessed the use of any specific source of treatment versus none. We were not able, for instance, to make distinctions between those who attended a single session of treatment and those who completed a full course of treatment. Last, the NESARC data do not make distinctions between discrete treatment episodes, so our analyses were limited to examining types of treatment that were received within a specific time period.

5. Conclusion This work represents an initial step towards measuring and conceptualizing executed treatment utilization patterns and elucidating how specific common patterns of treatment utilization may produce differential alcohol-related outcomes. The findings presented here have implications for designing secondary prevention programs and for informing the behavior of addiction treatment practitioners, especially when treating individuals new to treatment for alcohol-related problems. As discussed previously, there may be a need to educate addiction treatment practitioners and AA programs that crosssector treatment referrals may optimize treatment outcomes. Moreover, the continued emergence of non-specialty treatment providers (e.g., primary care providers) in providing alcohol-related care may call for an increased need for training in referral practices, which could include encouragement to refer alcohol-affected individuals to multiple sectors of care.

Please cite this article as: Mowbray, O., et al., Latent Class Analysis of Alcohol Treatment Utilization Patterns and 3-Year Alcohol Related Outcomes, Journal of Substance Abuse Treatment (2015), http://dx.doi.org/10.1016/j.jsat.2015.01.012

O. Mowbray et al. / Journal of Substance Abuse Treatment xxx (2015) xxx–xxx

Appendix A. Estimated posterior probabilities of obtaining the 13 types of treatment for alcohol use. These results are a sensitivity analysis that included NESARC respondents who had ever sought treatment for drinking in their lifetime based on reports at the baseline or follow-up interview (n = 1,247).

n = 1,247

Multiservice use (12.1%)

AA Alone (24.4%)

AA combined with specialty (34.9%)

Private professionals (28.6%)

AA FS Detox Inpatient Outpatient Rehab ER Halfway Crisis EAP Clergy Private Other

0.979 0.619 0.914 0.858 0.817 0.955 0.812 0.280 0.316 0.236 0.475 0.916 0.249

0.843 0.022 0.063 0.000 0.009 0.205 0.164 0.036 0.000 0.000 0.071 0.000 0.122

0.933 0.251 0.630 0.355 0.486 0.891 0.025 0.101 0.016 0.094 0.170 0.514 0.145

0.366 0.296 0.009 0.071 0.106 0.019 0.152 0.000 0.017 0.057 0.156 0.693 0.123

Note: AA = Alcoholics Anonymous. FS = family or social service settings. Detox = alcohol or drug detoxification ward or clinic. Inpatient = inpatient ward of psychiatric or general hospital or community mental health program. Outpatient = outpatient clinic, including outreach programs and day or partial patient programs. Rehab = alcohol or drug rehabilitation program ER = Emergency room. Halfway = halfway house or therapeutic community. Crisis = crisis center. EAP = employee assistance program. Clergy = clergyman, priest, rabbi, or any type of religious counselor. Private = private physician, psychiatrist, psychologist, social worker, or any other professional. Other = other agency or professional. Probabilities greater than 0.4 are displayed in bold to assist with interpretation.

References Akaike, H. (1974). A new look at the statistical model identification. IEEE Transactions on Automatic Control, 19(6), 716–723, http://dx.doi.org/10.1109/TAC.1974.1100705. Amaro, H., Reed, E., Rowe, E., Picci, J., Mantella, P., & Prado, G. (2010). Brief screening and intervention for alcohol and drug use in a college student health clinic: Feasibility, implementation, and outcomes. Journal of American College Health, 58(4), 357–364, http://dx.doi.org/10.1080/07448480903501764. Asparouhov, T. (2005). Sampling weights in latent variable modeling. Structural Equation Modeling: A Multidisciplinary Journal, 12(3), 411–434, http://dx.doi.org/10.1207/ s15328007sem1203_4. Bottlender, M., & Soyka, M. (2005). Efficacy of an intensive outpatient rehabilitation program in alcoholism: Predictors of outcome 6 months after treatment. European Addiction Research, 11(3), 132–137, http://dx.doi.org/10.1159/000085548. Bradley, K. A., DeBenedetti, A. F., Volk, R. J., Williams, E. C., Frank, D., & Kivlahan, D. R. (2007). AUDIT-C as a brief screen for alcohol misuse in primary care. Alcoholism: Clinical and Experimental Research, 31(7), 1208–1217, http://dx.doi.org/10.1111/j. 1530-0277.2007.00403.x. Bush, K., Kivlahan, D. R., McDonell, M. B., Fihn, S. D., & Bradley, K. A. (1998). The audit alcohol consumption questions (AUDIT-C): An effective brief screening test for problem drinking. Archives of Internal Medicine, 158(16), 1789–1795, http://dx.doi.org/ 10.1001/archinte.158.16.1789. Cohen, E., Feinn, R., Arias, A., & Kranzler, H. R. (2007). Alcohol treatment utilization: Findings from the National Epidemiologic Survey on Alcohol and Related Conditions. Drug and Alcohol Dependence, 86(2–3), 214–221, http://dx.doi.org/10.1016/j.drugalcdep. 2006.06.008. Collins, L. M., & Lanza, S. T. (2013). Latent class and latent transition analysis: With applications in the social, behavioral, and health sciences. John Wiley & Sons. Dawson, D. A., Goldstein, R. B., & Grant, B. F. (2007). Rates and correlates of relapse among individuals in remission from DSM-IV alcohol dependence: A 3-year follow-up. Alcoholism: Clinical and Experimental Research, 31(12), 2036–2045, http://dx.doi.org/ 10.1111/j.1530-0277.2007.00536.x. Dawson, D. A., Goldstein, R. B., & Grant, B. F. (2012). Factors associated with first utilization of different types of care for alcohol problems. Journal of Studies on Alcohol and Drugs, 73(4), 647–656. Dawson, D. A., Grant, B. F., Stinson, F. S., & Chou, P. S. (2006). Estimating the effect of helpseeking on achieving recovery from alcohol dependence. Addiction, 101(6), 824–834, http://dx.doi.org/10.1111/j.1360-0443.2006.01433.x. Dawson, D. A., Grant, B. F., Stinson, F. S., Chou, Huang, B., & Ruan, W. J. (2006). Recovery from DSM-IV alcohol dependence: United States, 2001-2002. Alcohol Research & Health, 29(2), 131–142. Dawson, D. A., Grant, B. F., Stinson, F. S., & Zhou, Y. (2005). Effectiveness of the derived alcohol use disorders identification test (AUDIT-C) in screening for alcohol use disorders and risk drinking in the U.S. general population. Alcoholism: Clinical and Experimental Research, 29 (5), 844–854, http://dx.doi.org/10.1097/01.ALC.0000164374.32229.A2.

7

Ferri, M., Amato, L., & Davoli, M. (2006). Alcoholics Anonymous and other 12-step programmes for alcohol dependence. In The Cochrane Collaboration (Ed.), Cochrane Database of Systematic ReviewsChichester, UK: John Wiley & Sons, Ltd (Retrieved from http://www.drugsandalcohol.ie/13566/). Fiorentine, R. (1999). After drug treatment: Are 12-Step programs effective in maintaining abstinence? The American Journal of Drug and Alcohol Abuse, 25(1), 93–116, http://dx. doi.org/10.1081/ADA-100101848. Glass, J. E., & Bucholz, K. K. (2011). Concordance between self-reports and archival records of physician visits: A case–control study comparing individuals with and without alcohol use disorders in the community. Drug and Alcohol Dependence, 116(1–3), 57–63, http://dx.doi.org/10.1016/j.drugalcdep.2010.11.021. Glass, J. E., Perron, B. E., Ilgen, M. A., Chermack, S. T., Ratliff, S., & Zivin, K. (2010). Prevalence and correlates of specialty substance use disorder treatment for Department of Veterans Affairs Healthcare System patients with high alcohol consumption. Drug and Alcohol Dependence, 112(1–2), 150–155, http://dx.doi.org/10.1016/j. drugalcdep.2010.06.003. Gossop, M., Harris, J., Best, D., Man, L. H., Manning, V., & Marshall, J. (2003). Is attendance at alcoholics anonymous meetings after inpatient treatment related to improved outcomes? A 6-month follow-up study. Alcohol and Alcoholism, 38(5), 421–426, http:// dx.doi.org/10.1093/alcalc/agg104. Gossop, M., Stewart, D., & Marsden, J. (2008). Attendance at Narcotics Anonymous and Alcoholics Anonymous meetings, frequency of attendance and substance use outcomes after residential treatment for drug dependence: A 5-year follow-up study. Addiction, 103(1), 119–125, http://dx.doi.org/10.1111/j.1360-0443.2007.02050.x. Grant, B. F., Kaplan, K., & Stinson, F. S. (2007). Source and accuracy statement: Wave 2 National Epidemiologic Survey on Alcohol and Related Conditions. Bethesda, M.D.: National Institute of Alcoholism and Alcohol Abuse. Grant, B. F., Moore, T. C., Shepard, J., & Kaplan, K. (2003). Source and accuracy statement: Wave 1 National Epidemiologic Survey on Alcohol and Related Conditions (NESARC). Bethesda, M.D.: National Institute of Alcoholism and Alcohol Abuse. Hasin, D. S., Stinson, F. S., Ogburn, E., & Grant, B. F. (2007). Prevalence, correlates, disability, and comorbidity of DSM-IV alcohol abuse and dependence in the United States: Results from the National Epidemiologic Survey on Alcohol and Related Conditions. Archives of General Psychiatry, 64(7), 830–842, http://dx.doi.org/10.1001/archpsyc. 64.7.830. Hatzenbuehler, M. L., Keyes, K. M., Narrow, W. E., Grant, B. F., & Hasin, D. S. (2008). Racial/ Ethnic disparities in service utilization for individuals with co-occurring mental health and substance use disorders in the general population. The Journal of Clinical Psychiatry, 69(7), 1112–1121. Hermansson, U., Helander, A., Brandt, L., Huss, A., & Rönnberg, S. (2010). Screening and brief intervention for risky alcohol consumption in the workplace: Results of a 1year randomized controlled study. Alcohol and Alcoholism, 45(3), 252–257, http:// dx.doi.org/10.1093/alcalc/agq021. Humphreys, K. (1997). Clinicians’ referral and matching of substance abuse patients to self-help groups after treatment. Psychiatric Services (Washington, D.C.), 48(11), 1445–1449. Ilgen, M. A., Price, A. M., Burnett-Zeigler, I., Perron, B., Islam, K., Bohnert, A. S. B., et al. (2011). Longitudinal predictors of addictions treatment utilization in treatmentnaïve adults with alcohol use disorders. Drug and Alcohol Dependence, 113(2–3), 215–221, http://dx.doi.org/10.1016/j.drugalcdep.2010.08.006. Kaplan, D., & Ferguson, A. J. (1999). On the utilization of sample weights in latent variable models. Structural Equation Modeling: A Multidisciplinary Journal, 6(4), 305–321, http://dx.doi.org/10.1080/10705519909540138. Kelly, J. F., Stout, R., Zywiak, W., & Schneider, R. (2006). A 3-year study of addiction mutual-help group participation following intensive outpatient treatment. Alcoholism: Clinical and Experimental Research, 30(8), 1381–1392, http://dx.doi.org/ 10.1111/j.1530-0277.2006.00165.x. Keso, L., & Salaspuro, M. (1990). Inpatient treatment of employed alcoholics: A randomized clinical trial on Hazelden-type and traditional treatment. Alcoholism: Clinical and Experimental Research, 14(4), 584–589, http://dx.doi.org/10.1111/j.1530-0277. 1990.tb01206.x. Kessler, R. C., Nelson, C. B., McGonagle, K. A., Edlund, M. J., Frank, R. G., & Leaf, P. J. (1996). The epidemiology of co-occurring addictive and mental disorders: Implications for prevention and service utilization. American Journal of Orthopsychiatry, 66(1), 17–31, http://dx.doi.org/10.1037/h0080151. Killeen, T. K., Brady, K. T., Gold, P. B., Tyson, C., & Simpson, K. N. (2004). Comparison of self-report versus agency records of service utilization in a community sample of individuals with alcohol use disorders. Drug and Alcohol Dependence, 73(2), 141–147, http://dx.doi.org/10.1016/j.drugalcdep.2003.09.006. Kownacki, R. J., & Shadish, W. R. (1999). Does Alcoholics Anonymous work? The results from a meta-analysis of controlled experiments. Substance Use & Misuse, 34(13), 1897–1916, http://dx.doi.org/10.3109/10826089909039431. Lo, Y., Mendell, N. R., & Rubin, D. B. (2001). Testing the number of components in a normal mixture. Biometrika, 88(3), 767–778, http://dx.doi.org/10.1093/biomet/88.3.767. Montgomery, H. A., Miller, W. R., & Scott Tonigan, J. (1995). Does alcoholics anonymous involvement predict treatment outcome? Journal of Substance Abuse Treatment, 12 (4), 241–246, http://dx.doi.org/10.1016/0740-5472(95)00018-Z. Mowbray, O. (2014). The moderating role of social networks in the relationship between alcohol consumption and treatment utilization for alcohol-related problems. Journal of Substance Abuse Treatment, 46(5), 597–601, http://dx.doi.org/ 10.1016/j.jsat.2013.12.001. Muthén, L. K., & Muthén, B. O. (2012). Mplus User’s Guide (7th ed.) (Los Angeles, CA). Nylund, K. L., Asparouhov, T., & Muthén, B. O. (2007). Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carlo simulation study. Structural Equation Modeling: A Multidisciplinary Journal, 14(4), 535–569, http://dx.doi.org/10.1080/10705510701575396.

Please cite this article as: Mowbray, O., et al., Latent Class Analysis of Alcohol Treatment Utilization Patterns and 3-Year Alcohol Related Outcomes, Journal of Substance Abuse Treatment (2015), http://dx.doi.org/10.1016/j.jsat.2015.01.012

8

O. Mowbray et al. / Journal of Substance Abuse Treatment xxx (2015) xxx–xxx

Ramaswamy, V., Desarbo, W. S., Reibstein, D. J., & Robinson, W. T. (1993). An empirical pooling approach for estimating marketing mix elasticities with PIMS data. Marketing Science, 12(1), 103–124, http://dx.doi.org/10.1287/mksc.12.1.103. Regier, D. A., Narrow, W. E., Rae, D. S., Manderscheid, R. W., Locke, B. Z., & Goodwin, F. K. (1993). The de facto US mental and addictive disorders service system. Epidemiologic catchment area prospective 1-year prevalence rates of disorders and services. Archives of General Psychiatry, 50(2), 85–94. Riordan, R. J., & Walsh, L. (1994). Guidelines for professional referral to Alcoholics Anonymous and other twelve step groups. Journal of Counseling & Development, 72(4), 351–355, http://dx.doi.org/10.1002/j.1556-6676.1994.tb00948.x. Roesch, S. C., Villodas, M., & Villodas, F. (2010). Latent class/profile analysis in maltreatment research: A commentary on Nooner et al., Pears et al., and looking beyond. Child Abuse & Neglect, 34(3), 155–160, http://dx.doi.org/10.1016/j.chiabu.2010.01.003. Ruan, W. J., Goldstein, R. B., Chou, S. P., Smith, S. M., Saha, T. D., Pickering, R. P., et al. (2008). The Alcohol Use Disorder and Associated Disabilities Interview Schedule-IV (AUDADISIV): Reliability of new psychiatric diagnostic modules and risk factors in a general population sample. Drug and Alcohol Dependence, 92(1–3), 27–36, http://dx.doi.org/10.1016/j. drugalcdep.2007.06.001. Saitz, R., Horton, N. J., Sullivan, L. M., Moskowitz, M. A., & Samet, J. H. (2003). Addressing alcohol problems in primary care: A cluster randomized, controlled trial of a systems intervention. The screening and intervention in primary care (SIP) study. Annals of Internal Medicine, 138(5), 372–382. Schonfeld, L., King-Kallimanis, B. L., Duchene, D. M., Etheridge, R. L., Herrera, J. R., Barry, K. L., et al. (2010). Screening and brief intervention for substance misuse among older adults: The Florida BRITE project. American Journal of Public Health, 100(1), 108–114, http://dx.doi.org/10.2105/AJPH.2008.149534.

Schwarz, G. (1978). Estimating the dimension of a model. The Annals of Statistics, 6(2), 461–464, http://dx.doi.org/10.1214/aos/1176344136. Sclove, S. L. (1987). Application of model-selection criteria to some problems in multivariate analysis. Psychometrika, 52(3), 333–343, http://dx.doi.org/10. 1007/BF02294360. StataCorp (2013). Stata Statistical Software: Release 13. College Station, TX: StataCorp LP. Substance Abuse and Mental Health Services Administration, Office of Applied Studies (2009). Treatment Episode Data Set (TEDS). Highlights - 2007. National Admissions to Substance Abuse Treatment Services. Rockville, MD: DASIS Series: S-45, DHHS Publication No. (SMA) 09-4360. Timko, C., Moos, R. H., Finney, J. W., & Lesar, M. D. (2000). Long-term outcomes of alcohol use disorders: Comparing untreated individuals with those in Alcoholics Anonymous and formal treatment. Journal of Studies on Alcohol and Drugs, 61(4), 529–540. Tonigan, J. S., Toscova, R., & Miller, W. R. (1996). Meta-analysis of the literature on Alcoholics Anonymous: Sample and study characteristics moderate findings. Journal of Studies on Alcohol and Drugs, 57(1), 65–72. Waisberg, J. L. (1990). Patient characteristics and outcome of inpatient treatment for alcoholism. Advances in Alcohol & Substance Abuse, 8(3–4), 9–32, http://dx.doi.org/10. 1300/J251v08n03_02. Weisner, C., Matzger, H., & Kaskutas, L. A. (2003). How important is treatment? One-year outcomes of treated and untreated alcohol-dependent individuals. Addiction, 98(7), 901–911, http://dx.doi.org/10.1046/j.1360-0443.2003.00438.x. Weisner, C., Mertens, J., Parthasarathy, S., Moore, Hunkeler, E. M., Hu, T., et al. (2000). The outcome and cost of alcohol and drug treatment in an HMO: Day hospital versus traditional outpatient regimens. Health Services Research, 35(4), 791–812.

Please cite this article as: Mowbray, O., et al., Latent Class Analysis of Alcohol Treatment Utilization Patterns and 3-Year Alcohol Related Outcomes, Journal of Substance Abuse Treatment (2015), http://dx.doi.org/10.1016/j.jsat.2015.01.012

Latent class analysis of alcohol treatment utilization patterns and 3-year alcohol related outcomes.

People who obtain treatment for alcohol use problems often utilize multiple sources of help. While prior studies have classified treatment use pattern...
326KB Sizes 0 Downloads 6 Views